The environmental impact of agricultural trade liberalization in Canada

Transcription

1 The environmental impact of agricultural trade liberalization in Canada Valeria C. Castellanos-Hurtado Faculty of Land and Food Systems The University of British Columbia

2 1. Motivation 2. Risk of nitrogen water contamination from farmland 3. Trade liberalization and environmental externalities from agriculture 4. Data sources and Estimations 5. Results and final remarks 2

3 Agriculture and the environment in Canada Environmental degradation from agriculture Degradation of downstream water courses: Nitrogen, phosphorus (fertilizer), coliforms (manure) Pesticides, herbicides and insecticides Focus on nitrogen More data available Most of it, in the form of nitrate, is soluble in water and can move into groundwater or surface water Harm to aquatic life, human health (drinking water), greenhouse gases 3

4 Residual soil nitrogen on farmland Agriculture and Agri-Food Canada, Agri-Environmental Indicators

5 Residual soil nitrogen on farmland Agriculture and Agri-Food Canada, Agri-Environmental Indicators

6 Residual soil nitrogen on farmland Agriculture and Agri-Food Canada, Agri-Environmental Indicators

7 Residual soil nitrogen on farmland Agriculture and Agri-Food Canada, Agri-Environmental Indicators

8 Residual soil nitrogen on farmland Agriculture and Agri-Food Canada, Agri-Environmental Indicators

9 Residual soil nitrogen on farmland Agriculture and Agri-Food Canada, Agri-Environmental Indicators

10 Environmental issues and agricultural trade Increased pollution around 1996, 2001, coinciding with Canadian major trade liberalization CUSTA in 1989, WTO 1995 How does increased pollution relate to agric. trade liberalization? Crop choices depend mainly on crop demand (i.e. prices), factor costs, soil type, weather type Assuming full trade liberalization, producers will switch to those crops for which they have a comparative advantage and that result in greater profit Different N requirements per crop 10

11 Agricultural trade liberalization in Canada Merchandise exports balance of payments volume index, annual ( 1971=100) CUSTA NAFTA (1994) WTO, AUSTRALIA (1995) ISRAEL, CHILE (1997) MERCOSUR* (1998) Exports of food, feed, beverages and tobacco Wheat, exports Other cereals unmilled, exports Other food, feed, beverages and tobacco, exports Other crude vegetable products, exports Barley, exports Wheat flour, exports Other cereal preparations, exports Rapeseed, exports Source: Statistics Canada. Tables ,0012,0018,0049, CANSIM (database). 11

12 Agricultural trade liberalization in Canada As predicted by most international trade models, Canadian agricultural exports increased considerably after trade liberalization Production volumes for most of the main crops increased as well The net effect on the environment, however, will not only depend on the scale of production, but also on the resulting crop composition Increase in the proportion of rapeseed (canola), maize, soybeans and peas Decrease in the proportion of barley and oats Proportion of wheat increased, then decreased Composition of maize for forage and silage decreased, then increased 12

13 Environmental issues and agricultural trade Agricultural trade liberalization Changes in relative prices within domestic economy Changes in production scale, composition, technique Increase/decrease in nitrogen pollution 13

14 Impact of trade liberalization on RSN in Canada 1. Econometric analysis of: Fertilizer use as a function of crops, to reflect crop N intake RSN as a function of commercial fertilizer, manure, agricultural techniques (summerfallow, tillage, irrigation) and farm/farm owner characteristics To determine which crops and farm/production characteristics are relevant to RSN levels; then: 2. Econometric analysis of trade liberalization as a determinant of crop production and fertilizer usage (still under work) 14

15 Estimations: Data 1. AAFC: Residual Soil Nitrogen indicator (kg N/ha) Estimates the residual N in the top 60 cm of soil at the end of the cropping season using Census of Agriculture data Census years: every five years Using GIS, superimposed the boundaries of CD to calculate RSN per CD. In areas where different levels arose, calculated weighted averages of RSN per polygon area 15

16 Estimations: Data 2. Census of Agriculture, Crop area (ha), fertilizer use, irrigation use, summerfallow area, livestock numbers (farms, head), farm characteristics Significant variation in Census Divisions (CD) across years, especially Adjusted with GIS software through aggregation of Census Consolidated Subdivisions to generate CD comparable across years 16

17 Estimations - Fertilizer use as a function of crops Fertilizer it = φ 1 + φ 2 crop area (%) ijt + u it ; i: CD, j: crop, t: year This estimation will show the relationship between the percentage of each aggregate crop group (i.e. crop composition), and the use of fertilizer Higher oilseed/beans/grains compositions should represent higher fertilizer use Low N-requiring crops such as pasture or fruit trees should have a negative relationship with fertilizer use Estimated first as a panel-level OLS regression No autocorrelation or multicollinearity Some heteroskedasticity, corrected through GLS panel estimation with no changes in equation/var efficiency 17

18 Fertilizer OLS, random GLS w/heterosk. OLS, fixed effects effects Correct. Grains (std err) ( ) ( ) ( ) (p-val) (0.000) (0.000) (0.000) Forage (std err) ( ) ( ) ( ) (p-val) (0.000) (0.000) (0.000) Pea/beans/ oilseeds (std err) ( ) ( ) ( ) (p-val) (0.000) (0.000) (0.000) Vegetables (std err) ( ) ( ) ( ) (p-val) (0.000) (0.294) (0.000) 18

19 Fertilizer OLS, random effects OLS, fixed effects GLS w/heterosk. Correct. Fruits and berries (std err) ( ) ( ) ( ) (p-val) (0.024) (0.726) (0.024) Greenhouse veg. and sod (std err) ( ) ( ) ( ) (p-val) (0.000) (0.000) (0.000) Potatoes, sugarbeets, other field cr. (std err) ( ) ( ) ( ) (p-val) (0.000) (0.000) (0.000) R2 within R2 between R2 overall Observations Groups

20 Estimations RSN as a function of fertilizer use and crop production tech. As expected, more nitrogen-demanding crops have a direct relationship with fertilizer use, while less nitrogen-demanding crops are either non-significant or have a negative relationship with the level of fertilizer in the period 20

21 Estimations RSN as a function of fertilizer use and crop production tech. Ideally, explain RSN in terms of fertilizer and manure inputs per crop since RSN depends on N input and crop N intake No time series on fertilizer/manure application per crop. Furthermore, total manure appl. not available in some census years Instead, analyze RSN in terms of predicted fertilizer use (from previous estimation) that reflects per-crop N intake, and cattle head per farm as a proxy for likelihood of manure use: RSN it = β 1 + β 2 fertilizerhat it + β 3 manureproxy it + β 4 irrigation it + β 5 summerfallow it +β 5 farm character it +ε it I expect a direct effect of fertilizer and manure (proxy) on RSN Uncertain effect from irrigation: improves N uptake in dry areas; overirrigation could lead to N leaching Summerfallow would have an inverse relationship with RSN as no fertilizer is needed 21

22 RSN OLS panel, fixed effects GLS, correct. Heterosk, autocorr Fertilizer (hat) (std err) (0.392) (0.242) (p-val) (0.114) (0.000) Cattle x farm (std err) (4.385) (1.891) (p-val) (0.000) (0.000) Irrigated ha as a percentage of total farmland Pasture (converted) ha as %of total farmland (34.91) (8.199) (0.55) (0.000) (15.12) (7.824) (0.000) (0.000) 22

23 RSN Pasture (natural) ha as % of total farmland Summerfallow ha as % of total farmland OLS panel, fixed effects GLS, correct. Heterosk, autocorr (9.225) (1.622) (0.000) (0.000) (17.56) (3.760) (0.000) (0.000) Observations Groups R2 within R2 between R2 overall Wald chi2(6) F(6,697) (0.000) (0.000) F test that all u_i=0: F(262, 697)

24 Estimations RSN as a function of fertilizer use and crop production tech. The relationship between RSN and N inputs is strongly supported: Having corrected for heteroskedasticity, there is a positive, significant relationship between (predicted) fertilizer and manure use, and RSN levels, as expected The presence of low N-requiring areas (pasture, summerfallow) has a negative, significant relationship with RSN Irrigation has a negative sign, suggesting less likelihood of over-irrigation; more likely irrigation occurs in dry areas and thus improves N intake Evidence of good land management practices 24

25 RSN GLS corrected heterosk, autocorr Fertilizer (hat) (std err) (p-val) (0.246) (0.002) Cattle x farm (std err) (p-val) Irrigated ha as a percentage of total farmland Pasture (converted) ha as %of total farmland Pasture (natural) ha as % of total farmland (2.008) (0.000) (8.937) (0.000) (7.484) (0.000) (1.973) (0.000) 25

26 RSN GLS corrected heterosk, autocorr Summerfallow ha as % of total farmland (4.747) (0.000) % of farms with farm owner/main operator younger than (4.863) (0.000) between 35 and 55 years old (4.737) (0.011) Total area rented or leased from others, as a % of total farmland (2.279) (0.066) Sole proprietorship (std err) (3.188) (p-val) (0.000) Family owned farm (std err) (5.205) (p-val) (0.000) Observations 966 Groups 263 Wald chi2(10)

27 RSN GLS corrected heterosk, autocorr Fertilizer (hat) (std err) (0.239) (p-val) (0.014) Bulls x farm (std err) (132.6) (p-val) (0.000) Milk cows x farm (std err) (7.168) (p-val) (0.000) Beef cows x farm (std err) (11.78) (p-val) (0.000) Following Oenema, Oudendag, Velthot (2007), we d expect dairy cows to have the strongest predictability power for manure use; in this regression, it seems as efficient as other cattle variables 27

28 (std err) (p-val) (std err) (p-val) RSN GLS corrected heterosk, autocorr Heifers x farm (13.21) (0.000) Steers x farm Irrigated ha as a percentage of total farmland (std err) (p-val) Pasture (converted) ha as %of total farmland (std err) (p-val) Pasture (natural) ha as % of total farmland (std err) (p-val) Summerfallow ha as % of total farmland (std err) (p-val) (8.507) (0.000) (9.137) (0.000) (8.702) (0.000) (2.249) (0.000) (3.911) (0.000) 28

29 Findings and conclusions Use of fertilizer is in line with the nutrient requirements outlined by OECD, FAO and the handbooks for most provinces Grains and oilseeds, which are the main crops in Canadian farmlands, are the most significative in fertilizer use Pasture, forage, fruit and berry trees have an inverse (sometimes insignificant) effect in the use of fertilizer RSN also in line with previous literature: Heavily dependant on commercial fertilizer and the proximity of (proxy for) manure production Pasture and summerfallow have a negative effect on RSN Positive sign for irrigation, i.e. mostly used in dry areas, therefore improving N intake by crops 29

30 Findings and conclusions Additionally, one of the main sources of residual nitrogen (manure) is highly correlated with intensive cattle farming; average cattle number per farm has increased in recent years Products such as wheat (during the sampled period, until 2006) might not have responded as much as other crops to trade liberalization, due to their respective boards The important effect to measure, however, is the response of production of other goods (which would react to world prices) in the total composition of crop/livestock production 30

31 Further steps To complete analysis of trade liberalization impact on crop and livestock production To extend environmental analysis with regional proxies of soil and weather characteristics, to identify which areas might be more at risk Cross these results with crop production and response to trade liberalization to identify key regions 31

32 Appendix Data sources and acknowledgements Data Sources and Acknowledgements: 2010 Agriculture and Agri-Food Canada. All rights reserved. Agri-Environmental Indicators Soil Landscapes of Canada v3.0 Fundamental Drainage Areas of Canada Census of Agriculture adapted from Agriculture and Agri-Food Canada and Statistics Canada, customized tabulations, Census of Agriculture, CGC Base 1996, 2001, 2006, Census of Agriculture Regular Base 1971, 1976, 1981, 1986, Department of Natural Resources Canada. All rights reserved. Atlas of Canada 1:1,000,000 National Frameworks Data click on Agri- Environmental Indicators 32

33 Appendix RSN The RSN indicator is calculated as the difference between all N inputs (fertilizer and manure addition, N fixation by leguminous plants, wet and dry atmospheric deposition) and all N outputs (N removal from the soil via crop uptake, plus N losses through volatilization of ammonia, N2O and N2 emissions). The RSN indicator provides an estimate of the amount of unused N that remains in the soil at the end of the cropping season. A model was derived to estimate the RSN indicator in agricultural regions across Canada on the basis of Soil Landscape of Canada (SLC) polygons (Yang et al., 2007). RSN is estimated for each year from 1981 to 2006 using annual data where available (e.g. yields and fertilizer sales) and by interpolating the census of agriculture data between census years (e.g. crop area and livestock number). 33

34 Appendix: Data for trade estimations 3. Statistics Canada: Production Special data product from Statistics Canada Yearly production, seeded area, harvested area, yields from 1976 to 2011 Data per province and Census Agricultural Region (CAR) Barley Canola Chick Peas Corn for Grain Dry Field Peas Durum wheat Fall rye Flaxseed Oats Soybeans Spring rye Summerfallow Sunflower seed Total Canary seed Total lentils Total mustard seed Total rye Total spring wheat Total wheat Triticale Winter wheat 34

35 Appendix: Data for trade estimations 4. Statistics Canada: Prices (quarterly indexes) Input price indexes, : total input index, plus general index for a) building, b) machinery, c) crop production, d) animal production, e) supplies and f) interest, and sub-indexes for elements within these six categories (53 indexes total) Agricultural production price indexes : general indexes for: cattle and calves, dairy, eggs, fruit, grains, hogs, oilseeds, potatoes, poultry, total crops, total livestock and animal products, vegetables excluding potatoes 35

36 Appendix: Data for trade estimations 4. Statistics Canada: Prices Price indexes for specific commodities: Barley (1982-) and Wheat (1985-) for the Canadian Wheat Board Canola (1985-) Ontario Wheat Board (1985-) Rye (1985-) 36

37 Appendix: Agricultural trade liberalization in Canada Crop production composition in Canada, main 15 products, 1970 Tomatoes 1% Linseed 3% Apples 1% Carrots and turnips 0% Sugar beet 2% Lentils 0% Peas, dry 0% Oats 13% Wheat 21% Soybeans 1% Potatoes 6% Rapeseed 4% Maize 6% Barley 21% Maize for forage and silage 21% 37 Source: FAOSTAT, Production: Crops FAO Statistics Division February 2012

38 Appendix: Agricultural trade liberalization in Canada Crop production composition in Canada, main 15 products, 1985 Sugar beet 1% Tomatoes 1% Peas, dry 0% Lentils 0% Oats 4% Linseed 1% Apples 1% Carrots and turnips 0% Soybeans 2% Potatoes 4% Barley 19% Wheat 37% Maize for forage and silage 14% Maize 11% Rapeseed 5% 38 Source: FAOSTAT, Production: Crops FAO Statistics Division February 2012

39 Appendix: Agricultural trade liberalization in Canada Crop production composition in Canada, main 15 products, 1996 Sugar beet 1% Tomatoes 1% Soybeans 3% Lentils 0% Peas, dry 1% Oats 6% Linseed 1% Apples 1% Carrots and turnips 0% Potatoes 5% Wheat 38% Barley 20% Maize for forage and silage 7% Maize 10% Rapeseed 6% 39 Source: FAOSTAT, Production: Crops FAO Statistics Division February 2012

40 Peas, dry 4% Appendix: Agricultural trade liberalization in Canada Crop production composition in Canada, main 15 products, 2005 Sugar beet 1% Tomatoes 1% Oats 4% Soybeans 4% Potatoes 5% Lentils 2% Linseed 1% Apples 1% Wheat 31% Carrots and turnips 0% Barley 14% Rapeseed 12% Maize for forage and silage 9% Maize 11% 40 Source: FAOSTAT, Production: Crops FAO Statistics Division February 2012